New AI tool discovers mutations linking bladder cancer to tobacco smoking – The Science Board

September 26, 2022 News Editor

Their study, published September 23 in the journal Cell Genomics, lays out how the artificial intelligence (AI) tool found patterns of mutations caused by carcinogens and other DNA-altering processes. They believe their work could help researchers identify what environmental factors, such as exposure to tobacco smoke and ultraviolet radiation, cause cancer in certain patients.

Environmental exposures alter DNA in a unique way generating a specific pattern of mutations called a mutational signature. While a mutational signature from tobacco smoking has been previously detected in lung cancer, that was not the case for bladder cancer — until now.

UC San Diego researchers say they have demonstrated that there is a mutational signature from tobacco smoking in bladder cancer that is different than the signature found in lung cancer. In addition, the research team contends that this signature is also found in normal bladder tissues of tobacco smokers who have not developed bladder cancer, while it was not found in the bladder tissues of nonsmokers.

“What this signature tells us is that certain mutations in your DNA are due to exposure to tobacco smoke,” co-first author Marcos Diaz-Gay, a postdoctoral researcher in the lab of Ludmil Alexandrov, professor of bioengineering and cellular and molecular medicine at UC San Diego, said in a statement. “It doesn’t necessarily mean that you have cancer. But the more you smoke, the more mutations accumulate in your cells, and the more you increase your risk for developing cancer.”

Alexandrov’s lab developed the next-generation machine-learning tool — which extracts mutational signatures directly from large amounts of genetic data — and used it to analyze 23,827 sequenced human cancers. What the tool discovered were four mutational signatures, including the one in bladder cancer tied to tobacco smoking, that had not been previously detected by other tools.

The researchers also matched their machine-learning tool against 13 existing bioinformatics tools, which were evaluated for their ability to extract mutational signatures from more than 80,000 synthetic cancer samples and outperformed them all. Their AI tool detected 20% to 50% more true positive signatures, with five times less false positive signatures.

“In bioinformatics, this is the first time that such a comprehensive benchmarking has been done on this scale for mutational signature extraction,” Diaz-Gay said.

Going forward, the research team wants to develop a web-based tool so that researchers can use and profile more patients.

“Right now, this tool requires bioinformatics expertise to run it,” Alexandrov said. “What we want is to create a user-friendly version on the web, where researchers can just drop in a patient’s mutations, and it immediately gives you the set of mutational signatures and what processes caused them.”